The long-term objective of this research is to understand how differences in patient-provider communication affect disparities in the care received by persons with HIV. This knowledge will facilitate the development of evidence based interventions, with both patient and providers, to eliminate these disparities. A variety of types of disparities in HIV care between whites and non-whites have been demonstrated, including what may be the most consequential for patients, disparities in rates of initiation of and adhering to HIV antiretroviral medications (ARVs). These disparities have many root causes, including care system access, differences in preferences or beliefs about the value of treatments, and bias against stigmatized groups such as injection drug users. An important source of these disparities may be differences in the nature and quality of patient-provider communication, a topic which has received little attention from HIV researchers. To fill this gap, we developed and validated a new method to code and analysis audiotapes of patient-provider interactions in HIV care, the General Medical Interaction Analysis System (GMIAS). A unique feature of this coding system is that it can characterize utterances in patient-provider dialog that are specifically related to ARV medication adherence. We propose to use GMIAS to analyze transcripts of 435 patient-provider visits for persons with HIV collected from 45 physicians practicing at 4 different care sites in Baltimore, New York, Detroit, and Portland as part of an existing study. Patients were sampled from participating physicians'practice such that approximately half were white and half were non-white, which facilitates within-physician comparisons of racial differences. Using the GMIAS coding system, we propose the following specific aims: 1) Compare general patterns of patient-provider communication between white and non-white patients, 2) Compare ARV related adherence dialog between white and non-white patients, 3) Determine, using variance components analysis, how variance in patient-centeredness is partitioned between patients and providers, and 4) Determine whether 1-year HIV outcomes (changes in viral loads and CD4 counts) differ between blacks and whites, and the extent to which differences in overall and adherence specific dialog quality explain differences in HIV outcomes. The relevance of this research to public health is that efforts to eliminate disparities in care related to race/ethnicity will require better data about the sources of those disparities. The results of this study will help us better understand how patient-provider dialog contributes to these disparities in HIV care, and provide a much needed evidence base for efforts to eliminate these disparities. Project Narrative While important disparities between the care received by whites and non-whites with HIV infection have been documented, including differences in rates of receipt of antiretroviral therapy, the source of these disparities is poorly understood. This project will use a new method to code and analyze audiotapes of provider communication to understand how patient-provider dialogue differs by race and ethnicity. A unique feature of this new coding method is that it permits detailed assessment of antiretroviral medication adherence.